# Notebook Config

import sys
sys.path.append('/content/drive/MyDrive/Arewa_capstone_project')

# import setup_env for environment setup
from helpers import setup_env

setup_env()

GPU available
Reusing cached mean and std

# Viewing different leafs from each wards

%matplotlib inline
from helpers import get_data_location
from dataloader import visualize_one_batch, load_data
import multiprocessing
data_dir = get_data_location()
# use get_data_loaders to get the data_loaders dictionary. Use a batch_size
# of 5, a validation size of 0.01 and num_workers=multiprocessing.cpu_count() (all CPUs)
data_loaders, dataset_sizes, class_names  =  load_data(data_dir, batch_size = 5, valid_size = 0.1, num_workers = multiprocessing.cpu_count())

visualize_one_batch(data_loaders)

Reusing cached mean and std
Dataset mean: tensor([0.5750, 0.6695, 0.5853]), std: tensor([0.2373, 0.1887, 0.3300])
Reusing cached mean and std

#printing the mean and std.
#printing the distribution of the whole datasets combines all the three WARDS (PlantVillage) DIKUMAR, KASAISA and KUKARETA

from custom_visualization import load_data

batch_size = 32

dataloaders, dataset_sizes, class_names = load_data(
    data_dir="/content/drive/MyDrive/Arewa_capstone_project/dataset",
    batch_size=batch_size
)

# Print sizes
print("\n📊 Dataset sizes:")
print(f"Training set:   {dataset_sizes['train']} samples")
print(f"Validation set: {dataset_sizes['val']} samples")
print(f"Test set:       {dataset_sizes['test']} samples")
print(f"\n🧾 Class names: {class_names}")

Reusing cached mean and std
Dataset mean: tensor([0.5750, 0.6695, 0.5853]), std: tensor([0.2373, 0.1887, 0.3300])

📊 Dataset sizes:
Training set:   516 samples
Validation set: 64 samples
Test set:       66 samples

🧾 Class names: ['INFECTED', 'HEALTHY']

#Visualizing data from infected and healthy leafs

%matplotlib inline
from helpers import get_data_location
from load import visualize_one_batch, load_data
import multiprocessing
data_dir = get_data_location(path = "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM")
# use get_data_loaders to get the data_loaders dictionary. Use a batch_size
# of 5, a validation size of 0.01 and num_workers=multiprocessing.cpu_count() (all CPUs)
data_loaders, dataset_sizes, class_names  =  load_data(data_dir, batch_size = 5, valid_size = 0.1, num_workers = multiprocessing.cpu_count())

visualize_one_batch(data_loaders)

Reusing cached mean and std
Dataset mean: tensor([0.5750, 0.6695, 0.5853]), std: tensor([0.2373, 0.1887, 0.3300])
Reusing cached mean and std

#Reusing mean and std cached

from helpers import compute_mean_and_std
mean, std = compute_mean_and_std()
print(f"Dataset mean: {mean}, std: {std}")

Reusing cached mean and std
Dataset mean: tensor([0.5750, 0.6695, 0.5853]), std: tensor([0.2373, 0.1887, 0.3300])

from train_transfer_learning import *
from vit_model import ViTBase
from resnet_model import ResNetBase
from efficientnet_model import EfficientNetB0
from mobilenetv3 import MobileNetV3


if __name__ == '__main__':
    datasets_paths = {
        "Dikumari": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM",
        "Kasaisa": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KASAISA WARD FARM",
        "Kukareta": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KUKARETA WARD FARM",
        "Combined": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
    }

    model_names = ["vit_b_16", "resnet18", "mobilenet_v3", "efficientnet_b0"]

    results = []

    for model_name in model_names:
        for label, path in datasets_paths.items():
            print(f"\n📦 Training {model_name} on {label} dataset")
            train_loader, val_loader, class_names = get_dataloaders(path)
            model = build_model(model_name, num_classes=len(class_names))
            run_name = f"{model_name}_{label}"
            best_acc = train_and_evaluate(model, train_loader, val_loader, model_name, label, run_name, results)
            print(f"🏁 Finished {run_name} | Best Val Acc: {best_acc:.2f}%\n")

    # Save results summary to CSV
    results_df = pd.DataFrame(results)
    results_df.to_csv(os.path.join(SAVE_DIR, "all_model_results.csv"), index=False)
    print("📊 Results saved to all_model_results.csv")

📦 Training vit_b_16 on Dikumari dataset
Reusing cached mean and std

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ViT_B_16_Weights.IMAGENET1K_V1`. You can also use `weights=ViT_B_16_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)
wandb: Currently logged in as: abubakaralamin100 (abubakaralamin100-gombe-state-university) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin

Epoch 1/25: 100%|██████████| 3/3 [00:16<00:00,  5.60s/it]

✅ Saved Best Model for vit_b_16_Dikumari with Val Acc: 50.00%

Epoch 2/25: 100%|██████████| 3/3 [00:04<00:00,  1.59s/it]

✅ Saved Best Model for vit_b_16_Dikumari with Val Acc: 60.00%

Epoch 3/25: 100%|██████████| 3/3 [00:05<00:00,  1.82s/it]
Epoch 4/25: 100%|██████████| 3/3 [00:04<00:00,  1.63s/it]

✅ Saved Best Model for vit_b_16_Dikumari with Val Acc: 70.00%

Epoch 5/25: 100%|██████████| 3/3 [00:05<00:00,  1.86s/it]
Epoch 6/25: 100%|██████████| 3/3 [00:04<00:00,  1.54s/it]
Epoch 7/25: 100%|██████████| 3/3 [00:05<00:00,  1.75s/it]
Epoch 8/25: 100%|██████████| 3/3 [00:04<00:00,  1.54s/it]
Epoch 9/25: 100%|██████████| 3/3 [00:05<00:00,  1.80s/it]
Epoch 10/25: 100%|██████████| 3/3 [00:04<00:00,  1.58s/it]
Epoch 11/25: 100%|██████████| 3/3 [00:04<00:00,  1.63s/it]
Epoch 12/25: 100%|██████████| 3/3 [00:05<00:00,  1.71s/it]
Epoch 13/25: 100%|██████████| 3/3 [00:04<00:00,  1.57s/it]
Epoch 14/25: 100%|██████████| 3/3 [00:05<00:00,  1.89s/it]
Epoch 15/25: 100%|██████████| 3/3 [00:04<00:00,  1.57s/it]
Epoch 16/25: 100%|██████████| 3/3 [00:05<00:00,  1.87s/it]
Epoch 17/25: 100%|██████████| 3/3 [00:04<00:00,  1.61s/it]
Epoch 18/25: 100%|██████████| 3/3 [00:05<00:00,  1.70s/it]
Epoch 19/25: 100%|██████████| 3/3 [00:05<00:00,  1.76s/it]
Epoch 20/25: 100%|██████████| 3/3 [00:04<00:00,  1.58s/it]
Epoch 21/25: 100%|██████████| 3/3 [00:05<00:00,  1.85s/it]
Epoch 22/25: 100%|██████████| 3/3 [00:04<00:00,  1.55s/it]
Epoch 23/25: 100%|██████████| 3/3 [00:05<00:00,  1.80s/it]
Epoch 24/25: 100%|██████████| 3/3 [00:04<00:00,  1.54s/it]
Epoch 25/25: 100%|██████████| 3/3 [00:04<00:00,  1.64s/it]

Final Metrics for vit_b_16_Dikumari → Acc: 0.60, F1: 0.60, AUC: 0.6483516483516484

import pandas as pd

first_training_results_summary = pd.read_csv("/content/drive/MyDrive/Arewa_capstone_project/all_model_results.csv")
first_training_results_summary

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# === Load your first-run results ===
df = pd.DataFrame({
    "Run": [
        "vit_b_16_Dikumari", "vit_b_16_Kasaisa", "vit_b_16_Kukareta", "vit_b_16_Combined",
        "resnet18_Dikumari", "resnet18_Kasaisa", "resnet18_Kukareta", "resnet18_Combined",
        "mobilenet_v3_Dikumari", "mobilenet_v3_Kasaisa", "mobilenet_v3_Kukareta", "mobilenet_v3_Combined",
        "efficientnet_b0_Dikumari", "efficientnet_b0_Kasaisa", "efficientnet_b0_Kukareta", "efficientnet_b0_Combined"
    ],
    "Model": [
        "vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
        "resnet18", "resnet18", "resnet18", "resnet18",
        "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
        "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"
    ],
    "Dataset": [
        "Dikumari", "Kasaisa", "Kukareta", "Combined",
        "Dikumari", "Kasaisa", "Kukareta", "Combined",
        "Dikumari", "Kasaisa", "Kukareta", "Combined",
        "Dikumari", "Kasaisa", "Kukareta", "Combined"
    ],
    "Accuracy": [
        0.60, 0.916667, 0.55, 0.8125,
        0.70, 0.833333, 0.75, 0.84375,
        0.60, 0.791667, 0.65, 0.84375,
        0.70, 0.833333, 0.80, 0.796875
    ],
    "F1 Score": [
        0.596, 0.916, 0.540, 0.812,
        0.697, 0.833, 0.744, 0.842,
        0.583, 0.788, 0.627, 0.841,
        0.670, 0.829, 0.762, 0.793
    ],
    "ROC AUC": [
        0.648, 0.938, 0.659, 0.893,
        0.875, 0.929, 0.792, 0.943,
        0.747, 0.859, 0.810, 0.899,
        0.780, 0.953, 0.821, 0.855
    ]
})

# === Helper to annotate bars
def annotate_bars(ax, values, fmt="%.2f", offset=0.01):
    for bar, val in zip(ax.patches, values):
        ax.text(
            bar.get_x() + bar.get_width() / 2,
            bar.get_height() + offset,
            fmt % val,
            ha="center", va="bottom", fontsize=7
        )

# === Plot
sns.set(style="whitegrid")
fig, axes = plt.subplots(1, 3, figsize=(20, 6), sharex=True)

# 📈 Accuracy
sns.barplot(data=df, x="Dataset", y="Accuracy", hue="Model", ax=axes[0], palette="tab10")
annotate_bars(axes[0], df["Accuracy"], fmt="%.2f", offset=0.01)
axes[0].set_title("Accuracy")
axes[0].set_ylim(0, 1.1)
axes[0].set_ylabel("Accuracy")

# 🎯 F1
sns.barplot(data=df, x="Dataset", y="F1 Score", hue="Model", ax=axes[1], palette="pastel")
annotate_bars(axes[1], df["F1 Score"], fmt="%.2f", offset=0.01)
axes[1].set_title("F1 Score")
axes[1].set_ylim(0, 1.1)
axes[1].set_ylabel("")

# 🚀 AUC
sns.barplot(data=df, x="Dataset", y="ROC AUC", hue="Model", ax=axes[2], palette="Set2")
annotate_bars(axes[2], df["ROC AUC"], fmt="%.2f", offset=0.01)
axes[2].set_title("ROC-AUC")
axes[2].set_ylim(0, 1.1)
axes[2].set_ylabel("")

# Cleanup
for ax in axes:
    ax.tick_params(axis='x', rotation=10)
    ax.get_legend().remove()

# Shared legend
handles, labels = axes[0].get_legend_handles_labels()
fig.legend(handles, labels, loc="lower center", ncol=len(labels), fontsize=10, title="Model")

plt.tight_layout(rect=[0, 0.08, 1, 1])
save_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/first_run_metrics_summary.png"
plt.savefig(save_path)
plt.show()

print(f"✅ First training summary plot saved to:\n{save_path}")

✅ First training summary plot saved to:
/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/first_run_metrics_summary.png

from test_models import *
run_tests()

❌ Model not found: /content/drive/MyDrive/Arewa_capstone_project/best_model_vit_base_Dikumari.pth
❌ Model not found: /content/drive/MyDrive/Arewa_capstone_project/best_model_vit_base_Kasaisa.pth
❌ Model not found: /content/drive/MyDrive/Arewa_capstone_project/best_model_vit_base_Kukareta.pth
❌ Model not found: /content/drive/MyDrive/Arewa_capstone_project/best_model_vit_base_Combined.pth

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested resnet18 on Dikumari → Acc: 0.53, F1: 0.53, AUC: 0.5247058823529411

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested resnet18 on Kasaisa → Acc: 0.55, F1: 0.55, AUC: 0.562912087912088

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested resnet18 on Kukareta → Acc: 0.42, F1: 0.42, AUC: 0.3822641509433962

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ResNet18_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet18_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested resnet18 on Combined → Acc: 0.50, F1: 0.50, AUC: 0.4695332080200501

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=MobileNet_V3_Large_Weights.IMAGENET1K_V1`. You can also use `weights=MobileNet_V3_Large_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

import os
import torch
from test_models import load_test_loader, load_model, evaluate_model

# === Setup ===
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model_name = "vit_base"
train_label = "Combined"  # The ward it was trained on
model_path = f"/content/drive/MyDrive/Arewa_capstone_project/best_model_{model_name}_{train_label}.pth"

# === Paths to all test datasets ===
datasets_paths =datasets_paths = {
        "Dikumari": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM",
        "Kasaisa": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KASAISA WARD FARM",
        "Kukareta": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KUKARETA WARD FARM",
        "Combined": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
    }

# === Test on all datasets using the ViT model ===
for test_label, test_path in datasets_paths.items():
    test_loader, class_names = load_test_loader(test_path)
    model = load_model(model_name, num_classes=len(class_names), weights_path=model_path)
    run_name = f"{model_name}_{train_label}_tested_on_{test_label}"

    acc, f1, auc = evaluate_model(model, test_loader, class_names, run_name)

    print(f"✅ Tested ViT ({train_label}) on {test_label} → "
          f"Accuracy = {acc:.2f}, F1 = {f1:.2f}, AUC = {auc if auc else 'N/A'}")

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ViT_B_16_Weights.IMAGENET1K_V1`. You can also use `weights=ViT_B_16_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested ViT (Combined) on Dikumari → Accuracy = 0.47, F1 = 0.38, AUC = 0.4443137254901961

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ViT_B_16_Weights.IMAGENET1K_V1`. You can also use `weights=ViT_B_16_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested ViT (Combined) on Kasaisa → Accuracy = 0.56, F1 = 0.54, AUC = 0.6225274725274726

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ViT_B_16_Weights.IMAGENET1K_V1`. You can also use `weights=ViT_B_16_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested ViT (Combined) on Kukareta → Accuracy = 0.59, F1 = 0.59, AUC = 0.5218867924528302

/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and may be removed in the future, please use 'weights' instead.
  warnings.warn(
/usr/local/lib/python3.11/dist-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and may be removed in the future. The current behavior is equivalent to passing `weights=ViT_B_16_Weights.IMAGENET1K_V1`. You can also use `weights=ViT_B_16_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

✅ Tested ViT (Combined) on Combined → Accuracy = 0.47, F1 = 0.42, AUC = 0.5118264411027569

first_test_results_summary = pd.read_csv("/content/drive/MyDrive/Arewa_capstone_project/test_model_results.csv")
first_test_results_summary

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Load the first test summary data
df_test_summary = pd.DataFrame({
    "Model": [
        "resnet18", "resnet18", "resnet18", "resnet18",
        "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
        "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"
    ],
    "Test Dataset": [
        "Dikumari", "Kasaisa", "Kukareta", "Combined",
        "Dikumari", "Kasaisa", "Kukareta", "Combined",
        "Dikumari", "Kasaisa", "Kukareta", "Combined"
    ],
    "Accuracy": [
        0.514851, 0.483607, 0.582524, 0.493750,
        0.514851, 0.434426, 0.533981, 0.521875,
        0.465347, 0.500000, 0.427184, 0.515625
    ],
    "F1 Score": [
        0.434852, 0.480781, 0.576780, 0.491346,
        0.511788, 0.427931, 0.531818, 0.516086,
        0.456340, 0.478084, 0.427184, 0.501963
    ],
    "ROC AUC": [
        0.469412, 0.503022, 0.552830, 0.502937,
        0.503137, 0.465110, 0.483396, 0.537085,
        0.432941, 0.588462, 0.446415, 0.497024
    ]
})

# Function to annotate bars
def annotate_bars(ax, values, fmt="%.2f", offset=0.01):
    for bar, val in zip(ax.patches, values):
        ax.text(
            bar.get_x() + bar.get_width() / 2,
            bar.get_height() + offset,
            fmt % val,
            ha="center", va="bottom", fontsize=7
        )

# Plotting
sns.set(style="whitegrid")
fig, axes = plt.subplots(1, 3, figsize=(20, 6), sharex=True)

# Accuracy
sns.barplot(data=df_test_summary, x="Test Dataset", y="Accuracy", hue="Model", ax=axes[0], palette="tab10")
annotate_bars(axes[0], df_test_summary["Accuracy"], fmt="%.2f", offset=0.01)
axes[0].set_title("Accuracy")
axes[0].set_ylim(0, 1.05)

# F1 Score
sns.barplot(data=df_test_summary, x="Test Dataset", y="F1 Score", hue="Model", ax=axes[1], palette="pastel")
annotate_bars(axes[1], df_test_summary["F1 Score"], fmt="%.2f", offset=0.01)
axes[1].set_title("F1 Score")
axes[1].set_ylim(0, 1.05)

# ROC AUC
sns.barplot(data=df_test_summary, x="Test Dataset", y="ROC AUC", hue="Model", ax=axes[2], palette="Set2")
annotate_bars(axes[2], df_test_summary["ROC AUC"], fmt="%.2f", offset=0.01)
axes[2].set_title("ROC-AUC")
axes[2].set_ylim(0, 1.05)

# Adjust layout and legend
for ax in axes:
    ax.tick_params(axis='x', rotation=10)
    ax.get_legend().remove()

# Shared legend
handles, labels = axes[0].get_legend_handles_labels()
fig.legend(handles, labels, loc="lower center", ncol=len(labels), fontsize=10, title="Model")

plt.tight_layout(rect=[0, 0.08, 1, 1])
save_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/first_run_metrics_summary.png"
plt.savefig(save_path)
plt.show()

print(f"✅ First training summary plot saved to:\n{save_path}")

✅ First training summary plot saved to:
/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/first_run_metrics_summary.png

#Regularized Model training

import os
from torch.utils.data import random_split
import pandas as pd
from train_regularized_1 import train_model

SAVE_DIR = "/content/drive/MyDrive/Arewa_capstone_project/test_save_models"

datasets_paths = {
    "Dikumari": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM",
    "Kasaisa": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KASAISA WARD FARM",
    "Kukareta": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KUKARETA WARD FARM",
    "VillagePlant": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
}

model_names = ["vit_b_16", "resnet18", "mobilenet_v3", "efficientnet_b0"]

results = []
if __name__ == '__main__':
    for model_name in model_names:
        for label, path in datasets_paths.items():
            print(f"\n📦 Training {model_name} on {label} dataset")
            model, best_acc = train_model(data_dir=path, model_name=model_name, save_dir=SAVE_DIR)
            print(f"\n🏁 Finished {model_name}_{label} | Best Val Acc: {best_acc:.2f}%")
            results.append({
                "Model": model_name,
                "Dataset": label,
                "Best Val Accuracy (%)": best_acc
            })

    results_df = pd.DataFrame(results)
    results_df.to_csv(os.path.join(SAVE_DIR, "test_all_model_results.csv"), index=False)
    print("\n📊 Results saved to all_model_results.csv")

📦 Training vit_b_16 on Dikumari dataset

wandb: Currently logged in as: abubakaralamin100 (abubakaralamin100-gombe-state-university) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin

Reusing cached mean and std

/usr/local/lib/python3.11/dist-packages/torch/optim/lr_scheduler.py:62: UserWarning: The verbose parameter is deprecated. Please use get_last_lr() to access the learning rate.
  warnings.warn(
Epoch 1/10 - Training: 100%|██████████| 5/5 [00:06<00:00,  1.39s/it]
Epoch 1/10 - Validation: 100%|██████████| 2/2 [00:00<00:00,  4.71it/s]

Epoch 1: Train Loss 0.7497, Train Acc 51.25%, Val Loss 0.6778, Val Acc 70.00%
✅ Best model saved.

Epoch 2/10 - Training: 100%|██████████| 5/5 [00:05<00:00,  1.19s/it]
Epoch 2/10 - Validation: 100%|██████████| 2/2 [00:00<00:00,  4.53it/s]

Epoch 2: Train Loss 0.5687, Train Acc 80.00%, Val Loss 0.6673, Val Acc 50.00%
✅ Best model saved.

Epoch 3/10 - Training: 100%|██████████| 5/5 [00:06<00:00,  1.36s/it]
Epoch 3/10 - Validation: 100%|██████████| 2/2 [00:00<00:00,  4.84it/s]

Epoch 3: Train Loss 0.4442, Train Acc 87.50%, Val Loss 0.7199, Val Acc 40.00%

Epoch 4/10 - Training: 100%|██████████| 5/5 [00:06<00:00,  1.22s/it]
Epoch 4/10 - Validation: 100%|██████████| 2/2 [00:00<00:00,  4.65it/s]

reularized_Training_summary_val_acc = pd.read_csv("/content/drive/MyDrive/Arewa_capstone_project/test_save_models/test_all_model_results.csv")
reularized_Training_summary_val_acc

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Regularized training results
df_val_acc = pd.DataFrame({
    "Model": [
        "vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
        "resnet18", "resnet18", "resnet18", "resnet18",
        "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
        "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"
    ],
    "Dataset": [
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant"
    ],
    "Best Val Accuracy (%)": [
        72.727273, 92.307692, 81.818182, 68.750000,
        72.727273, 69.230769, 90.909091, 87.500000,
        54.545455, 92.307692, 72.727273, 71.875000,
        81.818182, 92.307692, 90.909091, 81.250000
    ]
})

# Plot
sns.set(style="whitegrid")
plt.figure(figsize=(12, 6))
ax = sns.barplot(data=df_val_acc, x="Dataset", y="Best Val Accuracy (%)", hue="Model", palette="muted")

# Annotate values
for p in ax.patches:
    ax.annotate(f'{p.get_height():.1f}%', (p.get_x() + p.get_width() / 2., p.get_height()),
                ha='center', va='bottom', fontsize=8, color='black', xytext=(0, 5), textcoords='offset points')

plt.title("Best Validation Accuracy (%) per Model per Dataset (Regularized Training)")
plt.ylim(0, 105)
plt.legend(title="Model", loc="lower center", bbox_to_anchor=(0.5, -0.25), ncol=4)
plt.tight_layout()

val_plot_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/regularized_validation_accuracy.png"
plt.savefig(val_plot_path)
plt.show()

val_plot_path

'/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/regularized_validation_accuracy.png'

import os
import torch
import pandas as pd
from train_regularized_1 import get_dataloaders, ViTBase, ResNetBase, EfficientNetB0, MobileNetV3, DEVICE

SAVE_DIR = "/content/drive/MyDrive/Arewa_capstone_project/test_save_models"

datasets_paths = {
    "Dikumari": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM",
    "Kasaisa": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KASAISA WARD FARM",
    "Kukareta": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KUKARETA WARD FARM",
    "VillagePlant": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
}

model_names = ["vit_b_16", "resnet18", "mobilenet_v3", "efficientnet_b0"]

# Model map for easy loading
def get_model(model_name, num_classes):
    if model_name == "vit_b_16":
        return ViTBase(n_classes=num_classes).to(DEVICE)
    elif model_name == "resnet18":
        return ResNetBase(n_classes=num_classes).to(DEVICE)
    elif model_name == "efficientnet_b0":
        return EfficientNetB0(n_classes=num_classes).to(DEVICE)
    elif model_name == "mobilenet_v3":
        return MobileNetV3(n_classes=num_classes, dropout=0.3).to(DEVICE)
    else:
        raise ValueError(f"Model {model_name} is not supported.")

def evaluate(model, test_loader):
    model.eval()
    criterion = torch.nn.CrossEntropyLoss()
    test_loss, test_correct = 0.0, 0
    with torch.no_grad():
        for images, labels in test_loader:
            images, labels = images.to(DEVICE), labels.to(DEVICE)
            outputs = model(images)
            loss = criterion(outputs, labels)
            test_loss += loss.item() * images.size(0)
            _, preds = torch.max(outputs, 1)
            test_correct += (preds == labels).sum().item()
    test_loss /= len(test_loader.dataset)
    test_acc = 100 * test_correct / len(test_loader.dataset)
    return test_loss, test_acc

results = []

if __name__ == "__main__":
    for model_name in model_names:
        for label, path in datasets_paths.items():
            print(f"\n🔍 Testing {model_name} on {label} dataset")

            # Get dataloaders to access test set
            _, _, test_loader, class_names = get_dataloaders(path)
            num_classes = len(class_names)

            # Load model
            model = get_model(model_name, num_classes)
            model_path = os.path.join(SAVE_DIR, f"best_model_{model_name}.pth")
            if not os.path.exists(model_path):
                print(f"⚠️ Model weights not found at {model_path}")
                continue
            model.load_state_dict(torch.load(model_path, map_location=DEVICE))

            # Evaluate
            test_loss, test_acc = evaluate(model, test_loader)
            print(f"✅ {model_name}_{label}: Test Acc = {test_acc:.2f}%, Test Loss = {test_loss:.4f}")

            results.append({
                "Model": model_name,
                "Dataset": label,
                "Test Accuracy (%)": test_acc,
                "Test Loss": test_loss
            })

    # Save results
    results_df = pd.DataFrame(results)
    results_file = os.path.join(SAVE_DIR, "test_results_all_models.csv")
    results_df.to_csv(results_file, index=False)
    print(f"\n📄 All results saved to {results_file}")

🔍 Testing vit_b_16 on Dikumari dataset
Reusing cached mean and std
✅ vit_b_16_Dikumari: Test Acc = 81.82%, Test Loss = 0.3503

🔍 Testing vit_b_16 on Kasaisa dataset
Reusing cached mean and std
✅ vit_b_16_Kasaisa: Test Acc = 92.31%, Test Loss = 0.1582

🔍 Testing vit_b_16 on Kukareta dataset
Reusing cached mean and std
✅ vit_b_16_Kukareta: Test Acc = 100.00%, Test Loss = 0.0598

🔍 Testing vit_b_16 on VillagePlant dataset
Reusing cached mean and std
✅ vit_b_16_VillagePlant: Test Acc = 100.00%, Test Loss = 0.0714

🔍 Testing resnet18 on Dikumari dataset
Reusing cached mean and std
✅ resnet18_Dikumari: Test Acc = 90.91%, Test Loss = 0.3435

🔍 Testing resnet18 on Kasaisa dataset
Reusing cached mean and std
✅ resnet18_Kasaisa: Test Acc = 100.00%, Test Loss = 0.1178

🔍 Testing resnet18 on Kukareta dataset
Reusing cached mean and std
✅ resnet18_Kukareta: Test Acc = 100.00%, Test Loss = 0.2229

🔍 Testing resnet18 on VillagePlant dataset
Reusing cached mean and std
✅ resnet18_VillagePlant: Test Acc = 93.75%, Test Loss = 0.2060

🔍 Testing mobilenet_v3 on Dikumari dataset
Reusing cached mean and std
✅ mobilenet_v3_Dikumari: Test Acc = 81.82%, Test Loss = 0.3580

🔍 Testing mobilenet_v3 on Kasaisa dataset
Reusing cached mean and std
✅ mobilenet_v3_Kasaisa: Test Acc = 100.00%, Test Loss = 0.1015

🔍 Testing mobilenet_v3 on Kukareta dataset
Reusing cached mean and std
✅ mobilenet_v3_Kukareta: Test Acc = 81.82%, Test Loss = 0.6540

🔍 Testing mobilenet_v3 on VillagePlant dataset
Reusing cached mean and std
✅ mobilenet_v3_VillagePlant: Test Acc = 87.50%, Test Loss = 0.3235

🔍 Testing efficientnet_b0 on Dikumari dataset
Reusing cached mean and std
✅ efficientnet_b0_Dikumari: Test Acc = 100.00%, Test Loss = 0.1034

🔍 Testing efficientnet_b0 on Kasaisa dataset
Reusing cached mean and std
✅ efficientnet_b0_Kasaisa: Test Acc = 100.00%, Test Loss = 0.0965

🔍 Testing efficientnet_b0 on Kukareta dataset
Reusing cached mean and std
✅ efficientnet_b0_Kukareta: Test Acc = 100.00%, Test Loss = 0.1713

🔍 Testing efficientnet_b0 on VillagePlant dataset
Reusing cached mean and std
✅ efficientnet_b0_VillagePlant: Test Acc = 96.88%, Test Loss = 0.1468

📄 All results saved to /content/drive/MyDrive/Arewa_capstone_project/test_save_models/test_results_all_models.csv

test_results_all_models = pd.read_csv("/content/drive/MyDrive/Arewa_capstone_project/test_save_models/test_results_all_models.csv")
test_results_all_models

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Create dataframe for test accuracy
df_test_acc = pd.DataFrame({
    "Model": [
        "vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
        "resnet18", "resnet18", "resnet18", "resnet18",
        "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
        "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"
    ],
    "Dataset": [
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant",
        "Dikumari", "Kasaisa", "Kukareta", "VillagePlant"
    ],
    "Test Accuracy (%)": [
        81.818182, 92.307692, 100.0, 100.0,
        90.909091, 100.0, 100.0, 93.75,
        81.818182, 100.0, 81.818182, 87.5,
        100.0, 100.0, 100.0, 96.875
    ]
})

# Plot
sns.set(style="whitegrid")
plt.figure(figsize=(12, 6))
ax = sns.barplot(data=df_test_acc, x="Dataset", y="Test Accuracy (%)", hue="Model", palette="colorblind")

# Annotate values
for p in ax.patches:
    ax.annotate(f'{p.get_height():.1f}%', (p.get_x() + p.get_width() / 2., p.get_height()),
                ha='center', va='bottom', fontsize=8, color='black', xytext=(0, 5), textcoords='offset points')

plt.title("Test Accuracy (%) per Model per Dataset")
plt.ylim(0, 110)
plt.legend(title="Model", loc="lower center", bbox_to_anchor=(0.5, -0.25), ncol=4)
plt.tight_layout()

test_plot_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_accuracy_summary.png"
plt.savefig(test_plot_path)
plt.show()

test_plot_path

'/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_accuracy_summary.png'

!pip install pytorch-gradcam matplotlib seaborn scikit-learn grad-cam

Collecting pytorch-gradcam
  Downloading pytorch-gradcam-0.2.1.tar.gz (6.0 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 6.0/6.0 MB 49.5 MB/s eta 0:00:00
  Preparing metadata (setup.py) ... done
Requirement already satisfied: matplotlib in /usr/local/lib/python3.11/dist-packages (3.10.0)
Requirement already satisfied: seaborn in /usr/local/lib/python3.11/dist-packages (0.13.2)
Requirement already satisfied: scikit-learn in /usr/local/lib/python3.11/dist-packages (1.6.1)
Collecting grad-cam
  Downloading grad-cam-1.5.5.tar.gz (7.8 MB)
     ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 7.8/7.8 MB 85.7 MB/s eta 0:00:00
  Installing build dependencies ... done
  Getting requirements to build wheel ... done
  Preparing metadata (pyproject.toml) ... done
Requirement already satisfied: opencv-python in /usr/local/lib/python3.11/dist-packages (from pytorch-gradcam) (4.11.0.86)
Requirement already satisfied: numpy in /usr/local/lib/python3.11/dist-packages (from pytorch-gradcam) (2.0.2)
Requirement already satisfied: contourpy>=1.0.1 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (1.3.2)
Requirement already satisfied: cycler>=0.10 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (0.12.1)
Requirement already satisfied: fonttools>=4.22.0 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (4.58.0)
Requirement already satisfied: kiwisolver>=1.3.1 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (1.4.8)
Requirement already satisfied: packaging>=20.0 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (24.2)
Requirement already satisfied: pillow>=8 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (11.2.1)
Requirement already satisfied: pyparsing>=2.3.1 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (3.2.3)
Requirement already satisfied: python-dateutil>=2.7 in /usr/local/lib/python3.11/dist-packages (from matplotlib) (2.9.0.post0)
Requirement already satisfied: pandas>=1.2 in /usr/local/lib/python3.11/dist-packages (from seaborn) (2.2.2)
Requirement already satisfied: scipy>=1.6.0 in /usr/local/lib/python3.11/dist-packages (from scikit-learn) (1.15.3)
Requirement already satisfied: joblib>=1.2.0 in /usr/local/lib/python3.11/dist-packages (from scikit-learn) (1.5.0)
Requirement already satisfied: threadpoolctl>=3.1.0 in /usr/local/lib/python3.11/dist-packages (from scikit-learn) (3.6.0)
Requirement already satisfied: torch>=1.7.1 in /usr/local/lib/python3.11/dist-packages (from grad-cam) (2.6.0+cu124)
Requirement already satisfied: torchvision>=0.8.2 in /usr/local/lib/python3.11/dist-packages (from grad-cam) (0.21.0+cu124)
Collecting ttach (from grad-cam)
  Downloading ttach-0.0.3-py3-none-any.whl.metadata (5.2 kB)
Requirement already satisfied: tqdm in /usr/local/lib/python3.11/dist-packages (from grad-cam) (4.67.1)
Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.11/dist-packages (from pandas>=1.2->seaborn) (2025.2)
Requirement already satisfied: tzdata>=2022.7 in /usr/local/lib/python3.11/dist-packages (from pandas>=1.2->seaborn) (2025.2)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.11/dist-packages (from python-dateutil>=2.7->matplotlib) (1.17.0)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (3.18.0)
Requirement already satisfied: typing-extensions>=4.10.0 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (4.13.2)
Requirement already satisfied: networkx in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (3.4.2)
Requirement already satisfied: jinja2 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (3.1.6)
Requirement already satisfied: fsspec in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (2025.3.2)
Collecting nvidia-cuda-nvrtc-cu12==12.4.127 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cuda_nvrtc_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-runtime-cu12==12.4.127 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cuda_runtime_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cuda-cupti-cu12==12.4.127 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cuda_cupti_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cudnn-cu12==9.1.0.70 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cublas-cu12==12.4.5.8 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cublas_cu12-12.4.5.8-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cufft-cu12==11.2.1.3 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cufft_cu12-11.2.1.3-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-curand-cu12==10.3.5.147 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_curand_cu12-10.3.5.147-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Collecting nvidia-cusolver-cu12==11.6.1.9 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cusolver_cu12-11.6.1.9-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Collecting nvidia-cusparse-cu12==12.3.1.170 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_cusparse_cu12-12.3.1.170-py3-none-manylinux2014_x86_64.whl.metadata (1.6 kB)
Requirement already satisfied: nvidia-cusparselt-cu12==0.6.2 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (0.6.2)
Requirement already satisfied: nvidia-nccl-cu12==2.21.5 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (2.21.5)
Requirement already satisfied: nvidia-nvtx-cu12==12.4.127 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (12.4.127)
Collecting nvidia-nvjitlink-cu12==12.4.127 (from torch>=1.7.1->grad-cam)
  Downloading nvidia_nvjitlink_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl.metadata (1.5 kB)
Requirement already satisfied: triton==3.2.0 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (3.2.0)
Requirement already satisfied: sympy==1.13.1 in /usr/local/lib/python3.11/dist-packages (from torch>=1.7.1->grad-cam) (1.13.1)
Requirement already satisfied: mpmath<1.4,>=1.1.0 in /usr/local/lib/python3.11/dist-packages (from sympy==1.13.1->torch>=1.7.1->grad-cam) (1.3.0)
Requirement already satisfied: MarkupSafe>=2.0 in /usr/local/lib/python3.11/dist-packages (from jinja2->torch>=1.7.1->grad-cam) (3.0.2)
Downloading nvidia_cublas_cu12-12.4.5.8-py3-none-manylinux2014_x86_64.whl (363.4 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 363.4/363.4 MB 3.3 MB/s eta 0:00:00
Downloading nvidia_cuda_cupti_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl (13.8 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 13.8/13.8 MB 45.9 MB/s eta 0:00:00
Downloading nvidia_cuda_nvrtc_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl (24.6 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 24.6/24.6 MB 60.3 MB/s eta 0:00:00
Downloading nvidia_cuda_runtime_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl (883 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 883.7/883.7 kB 19.3 MB/s eta 0:00:00
Downloading nvidia_cudnn_cu12-9.1.0.70-py3-none-manylinux2014_x86_64.whl (664.8 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 664.8/664.8 MB 2.8 MB/s eta 0:00:00
Downloading nvidia_cufft_cu12-11.2.1.3-py3-none-manylinux2014_x86_64.whl (211.5 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 211.5/211.5 MB 5.6 MB/s eta 0:00:00
Downloading nvidia_curand_cu12-10.3.5.147-py3-none-manylinux2014_x86_64.whl (56.3 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 56.3/56.3 MB 15.0 MB/s eta 0:00:00
Downloading nvidia_cusolver_cu12-11.6.1.9-py3-none-manylinux2014_x86_64.whl (127.9 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 127.9/127.9 MB 7.8 MB/s eta 0:00:00
Downloading nvidia_cusparse_cu12-12.3.1.170-py3-none-manylinux2014_x86_64.whl (207.5 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 207.5/207.5 MB 6.0 MB/s eta 0:00:00
Downloading nvidia_nvjitlink_cu12-12.4.127-py3-none-manylinux2014_x86_64.whl (21.1 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 21.1/21.1 MB 91.6 MB/s eta 0:00:00
Downloading ttach-0.0.3-py3-none-any.whl (9.8 kB)
Building wheels for collected packages: pytorch-gradcam, grad-cam
  Building wheel for pytorch-gradcam (setup.py) ... done
  Created wheel for pytorch-gradcam: filename=pytorch_gradcam-0.2.1-py3-none-any.whl size=5249 sha256=57a2665e308b0ba527a1c8cee06e5e89e1a9ed48202586cb904a597a1bdb2971
  Stored in directory: /root/.cache/pip/wheels/20/d6/4e/1ac02c54faee563b4d06015456c6c5d85550dcec878740f37d
  Building wheel for grad-cam (pyproject.toml) ... done
  Created wheel for grad-cam: filename=grad_cam-1.5.5-py3-none-any.whl size=44284 sha256=7ed4d315ed47ed86fceb872385021dd7c0b0cdec724c69fc165280c3b137ba89
  Stored in directory: /root/.cache/pip/wheels/bc/52/78/893c3b94279ef238f43a9e89608af648de401b96415bebbd1f
Successfully built pytorch-gradcam grad-cam
Installing collected packages: ttach, nvidia-nvjitlink-cu12, nvidia-curand-cu12, nvidia-cufft-cu12, nvidia-cuda-runtime-cu12, nvidia-cuda-nvrtc-cu12, nvidia-cuda-cupti-cu12, nvidia-cublas-cu12, pytorch-gradcam, nvidia-cusparse-cu12, nvidia-cudnn-cu12, nvidia-cusolver-cu12, grad-cam
  Attempting uninstall: nvidia-nvjitlink-cu12
    Found existing installation: nvidia-nvjitlink-cu12 12.5.82
    Uninstalling nvidia-nvjitlink-cu12-12.5.82:
      Successfully uninstalled nvidia-nvjitlink-cu12-12.5.82
  Attempting uninstall: nvidia-curand-cu12
    Found existing installation: nvidia-curand-cu12 10.3.6.82
    Uninstalling nvidia-curand-cu12-10.3.6.82:
      Successfully uninstalled nvidia-curand-cu12-10.3.6.82
  Attempting uninstall: nvidia-cufft-cu12
    Found existing installation: nvidia-cufft-cu12 11.2.3.61
    Uninstalling nvidia-cufft-cu12-11.2.3.61:
      Successfully uninstalled nvidia-cufft-cu12-11.2.3.61
  Attempting uninstall: nvidia-cuda-runtime-cu12
    Found existing installation: nvidia-cuda-runtime-cu12 12.5.82
    Uninstalling nvidia-cuda-runtime-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-runtime-cu12-12.5.82
  Attempting uninstall: nvidia-cuda-nvrtc-cu12
    Found existing installation: nvidia-cuda-nvrtc-cu12 12.5.82
    Uninstalling nvidia-cuda-nvrtc-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-nvrtc-cu12-12.5.82
  Attempting uninstall: nvidia-cuda-cupti-cu12
    Found existing installation: nvidia-cuda-cupti-cu12 12.5.82
    Uninstalling nvidia-cuda-cupti-cu12-12.5.82:
      Successfully uninstalled nvidia-cuda-cupti-cu12-12.5.82
  Attempting uninstall: nvidia-cublas-cu12
    Found existing installation: nvidia-cublas-cu12 12.5.3.2
    Uninstalling nvidia-cublas-cu12-12.5.3.2:
      Successfully uninstalled nvidia-cublas-cu12-12.5.3.2
  Attempting uninstall: nvidia-cusparse-cu12
    Found existing installation: nvidia-cusparse-cu12 12.5.1.3
    Uninstalling nvidia-cusparse-cu12-12.5.1.3:
      Successfully uninstalled nvidia-cusparse-cu12-12.5.1.3
  Attempting uninstall: nvidia-cudnn-cu12
    Found existing installation: nvidia-cudnn-cu12 9.3.0.75
    Uninstalling nvidia-cudnn-cu12-9.3.0.75:
      Successfully uninstalled nvidia-cudnn-cu12-9.3.0.75
  Attempting uninstall: nvidia-cusolver-cu12
    Found existing installation: nvidia-cusolver-cu12 11.6.3.83
    Uninstalling nvidia-cusolver-cu12-11.6.3.83:
      Successfully uninstalled nvidia-cusolver-cu12-11.6.3.83
Successfully installed grad-cam-1.5.5 nvidia-cublas-cu12-12.4.5.8 nvidia-cuda-cupti-cu12-12.4.127 nvidia-cuda-nvrtc-cu12-12.4.127 nvidia-cuda-runtime-cu12-12.4.127 nvidia-cudnn-cu12-9.1.0.70 nvidia-cufft-cu12-11.2.1.3 nvidia-curand-cu12-10.3.5.147 nvidia-cusolver-cu12-11.6.1.9 nvidia-cusparse-cu12-12.3.1.170 nvidia-nvjitlink-cu12-12.4.127 pytorch-gradcam-0.2.1 ttach-0.0.3

import os
import torch
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.metrics import accuracy_score, f1_score, confusion_matrix, roc_auc_score
from pytorch_grad_cam import GradCAM
from pytorch_grad_cam.utils.image import show_cam_on_image
from torchvision.transforms.functional import to_pil_image
from train_regularized_1 import get_dataloaders, ViTBase, ResNetBase, EfficientNetB0, MobileNetV3, DEVICE

# Paths
SAVE_DIR = "/content/drive/MyDrive/Arewa_capstone_project/test_save_models"
OUTPUT_DIR = "/content/drive/MyDrive/Arewa_capstone_project/all_test_save_models"
os.makedirs(OUTPUT_DIR, exist_ok=True)

# Dataset Paths
datasets_paths = {
    "Dikumari": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/DIKUMARI WARD FARM",
    "Kasaisa": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KASAISA WARD FARM",
    "Kukareta": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/KUKARETA WARD FARM",
    "VillagePlant": "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
}

model_names = ["vit_b_16", "resnet18", "mobilenet_v3", "efficientnet_b0"]

def get_model(model_name, num_classes):
    if model_name == "vit_b_16":
        return ViTBase(n_classes=num_classes).to(DEVICE)
    elif model_name == "resnet18":
        return ResNetBase(n_classes=num_classes).to(DEVICE)
    elif model_name == "efficientnet_b0":
        return EfficientNetB0(n_classes=num_classes).to(DEVICE)
    elif model_name == "mobilenet_v3":
        return MobileNetV3(n_classes=num_classes, dropout=0.3).to(DEVICE)
    else:
        raise ValueError(f"Model {model_name} is not supported.")

def evaluate(model, dataloader, class_names, save_prefix):
    model.eval()
    y_true, y_pred, y_probs = [], [], []
    with torch.no_grad():
        for images, labels in dataloader:
            images, labels = images.to(DEVICE), labels.to(DEVICE)
            outputs = model(images)
            probs = torch.softmax(outputs, dim=1)
            preds = torch.argmax(probs, dim=1)

            y_true.extend(labels.cpu().numpy())
            y_pred.extend(preds.cpu().numpy())
            y_probs.extend(probs[:, 1].detach().cpu().numpy())  # assumes binary classification

    acc = accuracy_score(y_true, y_pred)
    f1 = f1_score(y_true, y_pred)
    roc = roc_auc_score(y_true, y_probs)

    cm = confusion_matrix(y_true, y_pred)
    plt.figure(figsize=(5, 4))
    sns.heatmap(cm, annot=True, fmt='d', xticklabels=class_names, yticklabels=class_names, cmap='Blues')
    plt.title(f"Confusion Matrix - {save_prefix}")
    plt.xlabel("Predicted")
    plt.ylabel("True")
    plt.tight_layout()
    plt.savefig(os.path.join(OUTPUT_DIR, f"confusion_matrix_{save_prefix}.png"))
    plt.close()

    return acc, f1, roc

def generate_gradcam(model, dataloader, target_layer, save_prefix):
    model.eval()
    cam = GradCAM(model=model, target_layers=[target_layer], use_cuda=(DEVICE=="cuda"))

    images, labels = next(iter(dataloader))
    images = images.to(DEVICE)
    targets = [torch.nn.functional.one_hot(labels[i], num_classes=len(labels)).float().to(DEVICE) for i in range(len(labels))]

    grayscale_cam = cam(input_tensor=images, targets=None)
    for i, img in enumerate(images):
        rgb_img = img.permute(1, 2, 0).cpu().numpy()
        rgb_img = (rgb_img - rgb_img.min()) / (rgb_img.max() - rgb_img.min())
        cam_image = show_cam_on_image(rgb_img, grayscale_cam[i], use_rgb=True)

        cam_path = os.path.join(OUTPUT_DIR, f"{save_prefix}_gradcam_{i}.jpg")
        plt.imsave(cam_path, cam_image)

results = []

if __name__ == "__main__":
    for model_name in model_names:
        for label, path in datasets_paths.items():
            print(f"\n🔍 Testing {model_name} on {label} dataset")
            _, _, test_loader, class_names = get_dataloaders(path)
            num_classes = len(class_names)

            model = get_model(model_name, num_classes)
            model_path = os.path.join(SAVE_DIR, f"best_model_{model_name}.pth")
            if not os.path.exists(model_path):
                print(f"⚠️ Model weights not found: {model_path}")
                continue
            model.load_state_dict(torch.load(model_path, map_location=DEVICE))

            acc, f1, roc = evaluate(model, test_loader, class_names, f"{model_name}_{label}")
            print(f"✅ {model_name} on {label} -> Acc: {acc:.2%}, F1: {f1:.2f}, ROC-AUC: {roc:.2f}")

            # Choose last layer for GradCAM
            try:
                target_layer = list(model.children())[-1]
                generate_gradcam(model, test_loader, target_layer, f"{model_name}_{label}")
            except Exception as e:
                print(f"⚠️ Grad-CAM failed for {model_name}_{label}: {e}")

            results.append({
                "Model": model_name,
                "Dataset": label,
                "Test Accuracy (%)": acc * 100,
                "F1 Score": f1,
                "ROC-AUC": roc
            })

    df = pd.DataFrame(results)
    results_file = os.path.join(OUTPUT_DIR, "test_results_with_metrics.csv")
    df.to_csv(results_file, index=False)
    print(f"\n📄 All results saved to {results_file}")

🔍 Testing vit_b_16 on Dikumari dataset
Reusing cached mean and std

Exception ignored in: <function BaseCAM.__del__ at 0x7d412a1fd260>
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/dist-packages/pytorch_grad_cam/base_cam.py", line 212, in __del__
    self.activations_and_grads.release()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'GradCAM' object has no attribute 'activations_and_grads'

✅ vit_b_16 on Dikumari -> Acc: 90.91%, F1: 0.89, ROC-AUC: 1.00
⚠️ Grad-CAM failed for vit_b_16_Dikumari: GradCAM.__init__() got an unexpected keyword argument 'use_cuda'

🔍 Testing vit_b_16 on Kasaisa dataset
Reusing cached mean and std

Exception ignored in: <function BaseCAM.__del__ at 0x7d412a1fd260>
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/dist-packages/pytorch_grad_cam/base_cam.py", line 212, in __del__
    self.activations_and_grads.release()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'GradCAM' object has no attribute 'activations_and_grads'

✅ vit_b_16 on Kasaisa -> Acc: 100.00%, F1: 1.00, ROC-AUC: 1.00
⚠️ Grad-CAM failed for vit_b_16_Kasaisa: GradCAM.__init__() got an unexpected keyword argument 'use_cuda'

🔍 Testing vit_b_16 on Kukareta dataset
Reusing cached mean and std

Exception ignored in: <function BaseCAM.__del__ at 0x7d412a1fd260>
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/dist-packages/pytorch_grad_cam/base_cam.py", line 212, in __del__
    self.activations_and_grads.release()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'GradCAM' object has no attribute 'activations_and_grads'

✅ vit_b_16 on Kukareta -> Acc: 90.91%, F1: 0.89, ROC-AUC: 1.00
⚠️ Grad-CAM failed for vit_b_16_Kukareta: GradCAM.__init__() got an unexpected keyword argument 'use_cuda'

🔍 Testing vit_b_16 on VillagePlant dataset
Reusing cached mean and std

Exception ignored in: <function BaseCAM.__del__ at 0x7d412a1fd260>
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/dist-packages/pytorch_grad_cam/base_cam.py", line 212, in __del__
    self.activations_and_grads.release()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'GradCAM' object has no attribute 'activations_and_grads'

✅ vit_b_16 on VillagePlant -> Acc: 96.88%, F1: 0.97, ROC-AUC: 1.00
⚠️ Grad-CAM failed for vit_b_16_VillagePlant: GradCAM.__init__() got an unexpected keyword argument 'use_cuda'

🔍 Testing resnet18 on Dikumari dataset
Reusing cached mean and std

Downloading: "https://download.pytorch.org/models/resnet18-f37072fd.pth" to /root/.cache/torch/hub/checkpoints/resnet18-f37072fd.pth
100%|██████████| 44.7M/44.7M [00:00<00:00, 176MB/s]
Exception ignored in: <function BaseCAM.__del__ at 0x7d412a1fd260>
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/dist-packages/pytorch_grad_cam/base_cam.py", line 212, in __del__
    self.activations_and_grads.release()
    ^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'GradCAM' object has no attribute 'activations_and_grads'

test_all_model_results = pd.read_csv("/content/drive/MyDrive/Arewa_capstone_project/all_test_save_models/test_results_with_metrics.csv")
test_all_model_results

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# ========== Your Data ==========
df = pd.DataFrame({
    "Model": ["vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
              "resnet18", "resnet18", "resnet18", "resnet18",
              "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
              "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"],
    "Dataset": ["Dikumari", "Kasaisa", "Kukareta", "VillagePlant"] * 4,
    "Test Accuracy (%)": [
        90.91, 100.0, 90.91, 96.88,
        81.82, 100.0, 90.91, 90.63,
        90.91, 100.0, 72.73, 96.88,
        100.0, 100.0, 90.91, 96.88
    ],
    "F1 Score": [
        0.89, 1.00, 0.89, 0.97,
        0.75, 1.00, 0.91, 0.91,
        0.92, 1.00, 0.73, 0.96,
        1.00, 1.00, 0.92, 0.97
    ],
    "ROC-AUC": [
        1.00, 1.00, 1.00, 0.996,
        0.89, 1.00, 1.00, 1.00,
        0.96, 1.00, 0.96, 1.00,
        1.00, 1.00, 1.00, 0.967
    ]
})

# ========== Annotation Helper ==========
def annotate_bars(ax, values, fmt="%.2f", offset=1.0):
    for bar, val in zip(ax.patches, values):
        ax.text(
            bar.get_x() + bar.get_width() / 2,
            bar.get_height() + offset,
            fmt % val,
            ha="center", va="bottom", fontsize=7
        )

# ========== Plot ==========
sns.set(style="whitegrid")
fig, axes = plt.subplots(1, 3, figsize=(20, 6), sharex=True)

# 📈 Accuracy
sns.barplot(data=df, x="Dataset", y="Test Accuracy (%)", hue="Model", ax=axes[0], palette="tab10")
annotate_bars(axes[0], df["Test Accuracy (%)"], fmt="%.1f", offset=1.0)
axes[0].set_title("Test Accuracy (%)")
axes[0].set_ylim(0, 110)
axes[0].set_ylabel("Accuracy")

# 🎯 F1 Score
sns.barplot(data=df, x="Dataset", y="F1 Score", hue="Model", ax=axes[1], palette="pastel")
annotate_bars(axes[1], df["F1 Score"], fmt="%.2f", offset=0.03)
axes[1].set_title("F1 Score")
axes[1].set_ylim(0, 1.15)
axes[1].set_ylabel("F1")

# 🚀 ROC-AUC
sns.barplot(data=df, x="Dataset", y="ROC-AUC", hue="Model", ax=axes[2], palette="Set2")
annotate_bars(axes[2], df["ROC-AUC"], fmt="%.2f", offset=0.03)
axes[2].set_title("ROC-AUC Score")
axes[2].set_ylim(0, 1.15)
axes[2].set_ylabel("AUC")

# Clean up axes
for ax in axes:
    ax.tick_params(axis='x', rotation=10)
    ax.get_legend().remove()

# Shared Legend
handles, labels = axes[0].get_legend_handles_labels()
fig.legend(handles, labels, loc="lower center", ncol=len(labels), fontsize=10, title="Model")

# Layout & Save
plt.tight_layout(rect=[0, 0.08, 1, 1])
save_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_metrics_summary_plot.png"
plt.savefig(save_path)
plt.show()

print(f"✅ Metrics plot saved to:\n{save_path}")

✅ Metrics plot saved to:
/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_metrics_summary_plot.png

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Your DataFrame (same as before)
df = pd.DataFrame({
    "Model": ["vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
              "resnet18", "resnet18", "resnet18", "resnet18",
              "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
              "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"],
    "Dataset": ["Dikumari", "Kasaisa", "Kukareta", "VillagePlant"] * 4,
    "Test Accuracy (%)": [
        81.82, 92.31, 100.0, 100.0,
        90.91, 100.0, 100.0, 93.75,
        81.82, 100.0, 81.82, 87.5,
        100.0, 100.0, 100.0, 96.88
    ],
    "Test Loss": [
        0.3503, 0.1582, 0.0598, 0.0714,
        0.3435, 0.1178, 0.2229, 0.2060,
        0.3580, 0.1015, 0.6541, 0.3235,
        0.1034, 0.0965, 0.1713, 0.1468
    ]
})

# Helper to annotate bars without overlap
def annotate_bars(ax, fmt="%.1f%%", y_offset=1.5):
    for container in ax.containers:
        for bar in container:
            height = bar.get_height()
            ax.text(
                bar.get_x() + bar.get_width() / 2,
                height + y_offset,
                fmt % height,
                ha="center", va="bottom", fontsize=8
            )

# === Accuracy Plot ===
plt.figure(figsize=(12, 6))
ax = sns.barplot(data=df, x="Dataset", y="Test Accuracy (%)", hue="Model", palette="tab10")
annotate_bars(ax, fmt="%.1f%%", y_offset=1.5)
plt.title("📈 Test Accuracy per Model on Each Dataset")
plt.ylim(0, 110)
plt.ylabel("Accuracy (%)")
plt.xticks(rotation=10)
plt.legend(title="Model")
plt.tight_layout()
plt.savefig("/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_accuracy_comparison_fixed.png")
plt.show()

# === Loss Plot ===
plt.figure(figsize=(12, 6))
ax = sns.barplot(data=df, x="Dataset", y="Test Loss", hue="Model", palette="Set2")
annotate_bars(ax, fmt="%.2f", y_offset=0.01)
plt.title("📉 Test Loss per Model on Each Dataset")
plt.ylabel("Loss")
plt.xticks(rotation=10)
plt.legend(title="Model")
plt.tight_layout()
plt.savefig("/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_loss_comparison_fixed.png")
plt.show()

<ipython-input-9-ce6ad13034bb>:47: UserWarning: Glyph 128200 (\N{CHART WITH UPWARDS TREND}) missing from font(s) DejaVu Sans.
  plt.tight_layout()
<ipython-input-9-ce6ad13034bb>:48: UserWarning: Glyph 128200 (\N{CHART WITH UPWARDS TREND}) missing from font(s) DejaVu Sans.
  plt.savefig("/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_accuracy_comparison_fixed.png")
/usr/local/lib/python3.11/dist-packages/IPython/core/pylabtools.py:151: UserWarning: Glyph 128200 (\N{CHART WITH UPWARDS TREND}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)

<ipython-input-9-ce6ad13034bb>:59: UserWarning: Glyph 128201 (\N{CHART WITH DOWNWARDS TREND}) missing from font(s) DejaVu Sans.
  plt.tight_layout()
<ipython-input-9-ce6ad13034bb>:60: UserWarning: Glyph 128201 (\N{CHART WITH DOWNWARDS TREND}) missing from font(s) DejaVu Sans.
  plt.savefig("/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_loss_comparison_fixed.png")
/usr/local/lib/python3.11/dist-packages/IPython/core/pylabtools.py:151: UserWarning: Glyph 128201 (\N{CHART WITH DOWNWARDS TREND}) missing from font(s) DejaVu Sans.
  fig.canvas.print_figure(bytes_io, **kw)

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

# Your results data
df = pd.DataFrame({
    "Model": ["vit_b_16", "vit_b_16", "vit_b_16", "vit_b_16",
              "resnet18", "resnet18", "resnet18", "resnet18",
              "mobilenet_v3", "mobilenet_v3", "mobilenet_v3", "mobilenet_v3",
              "efficientnet_b0", "efficientnet_b0", "efficientnet_b0", "efficientnet_b0"],
    "Dataset": ["Dikumari", "Kasaisa", "Kukareta", "VillagePlant"] * 4,
    "Test Accuracy (%)": [
        81.82, 92.31, 100.0, 100.0,
        90.91, 100.0, 100.0, 93.75,
        81.82, 100.0, 81.82, 87.5,
        100.0, 100.0, 100.0, 96.88
    ],
    "Test Loss": [
        0.3503, 0.1582, 0.0598, 0.0714,
        0.3435, 0.1178, 0.2229, 0.2060,
        0.3580, 0.1015, 0.6541, 0.3235,
        0.1034, 0.0965, 0.1713, 0.1468
    ]
})

# Annotate helper
def annotate_bars(ax, values, fmt="%.1f%%", y_offset=1.5):
    for bar, val in zip(ax.patches, values):
        height = bar.get_height()
        ax.text(
            bar.get_x() + bar.get_width() / 2,
            height + y_offset,
            fmt % val,
            ha='center', va='bottom', fontsize=7
        )

# Plot
fig, axes = plt.subplots(1, 2, figsize=(16, 6), sharex=False)

# Accuracy subplot
sns.barplot(data=df, x="Dataset", y="Test Accuracy (%)", hue="Model", ax=axes[0], palette="tab10")
axes[0].set_title("Test Accuracy per Model")
axes[0].set_ylim(0, 110)
axes[0].set_ylabel("Accuracy (%)")
annotate_bars(axes[0], df["Test Accuracy (%)"].values, fmt="%.1f%%", y_offset=1.5)
axes[0].legend(title="Model", loc="lower right")

# Loss subplot
sns.barplot(data=df, x="Dataset", y="Test Loss", hue="Model", ax=axes[1], palette="Set2")
axes[1].set_title("Test Loss per Model")
axes[1].set_ylabel("Loss")
annotate_bars(axes[1], df["Test Loss"].values, fmt="%.2f", y_offset=0.01)
axes[1].legend(title="Model", loc="upper right")

# Layout and Save
plt.tight_layout()
combined_plot_path = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/test_accuracy_loss_combined.png"
plt.savefig(combined_plot_path)
plt.show()

print(f"✅ Combined plot saved to: {combined_plot_path}")

import os
import torch
import matplotlib.pyplot as plt
from torchvision import transforms
from train_regularized_1 import get_dataloaders, ViTBase, DEVICE

# Paths
dataset_path = "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
model_path = "/content/drive/MyDrive/Arewa_capstone_project/test_save_models/best_model_vit_b_16.pth"
output_file = "/content/vit_villageplant_predictions.png"

# Load data and model
_, _, test_loader, class_names = get_dataloaders(dataset_path)
model = ViTBase(n_classes=len(class_names)).to(DEVICE)
model.load_state_dict(torch.load(model_path, map_location=DEVICE))
model.eval()

# Unnormalize for display
inv_normalize = transforms.Normalize(
    mean=[-0.485 / 0.229, -0.456 / 0.224, -0.406 / 0.225],
    std=[1 / 0.229, 1 / 0.224, 1 / 0.225]
)

# Collect a batch for visualization
images, labels = next(iter(test_loader))
images, labels = images.to(DEVICE), labels.to(DEVICE)
with torch.no_grad():
    outputs = model(images)
    _, preds = torch.max(outputs, 1)
    corrects = preds == labels

# Plot
plt.figure(figsize=(20, 10))
for i in range(min(32, len(images))):
    img = inv_normalize(images[i].cpu()).clamp(0, 1)
    label = class_names[labels[i].item()]
    pred = class_names[preds[i].item()]
    correct = corrects[i].item()

    plt.subplot(4, 8, i + 1)
    plt.imshow(img.permute(1, 2, 0))
    plt.axis('off')
    color = 'green' if correct else 'red'
    plt.title(f"Pred: {pred}", color=color, fontsize=8)
    plt.xlabel(f"True: {label}", fontsize=8)

plt.tight_layout()
plt.savefig(output_file)
print(f"✅ Plot saved to: {output_file}")

Reusing cached mean and std

Downloading: "https://download.pytorch.org/models/vit_b_16-c867db91.pth" to /root/.cache/torch/hub/checkpoints/vit_b_16-c867db91.pth
100%|██████████| 330M/330M [00:02<00:00, 117MB/s]

✅ Plot saved to: /content/vit_villageplant_predictions.png

import os
import torch
import numpy as np
import matplotlib.pyplot as plt
from torchvision import transforms
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay
import os
import torch
import matplotlib.pyplot as plt
from torchvision import transforms
from train_regularized_1 import get_dataloaders, ViTBase, DEVICE
from pytorch_grad_cam import GradCAM
from pytorch_grad_cam.utils.model_targets import ClassifierOutputTarget
from pytorch_grad_cam.utils.image import show_cam_on_image


# Paths
dataset_path = "/content/drive/MyDrive/Arewa_capstone_project/Tuta absoluta tomato dataset/VillagePlant"
model_path = "/content/drive/MyDrive/Arewa_capstone_project/test_save_models/best_model_vit_b_16.pth"
output_file = "/content/vit_villageplant_predictions.png"

# Load data and model
_, _, test_loader, class_names = get_dataloaders(dataset_path)
model = ViTBase(n_classes=len(class_names)).to(DEVICE)
model.load_state_dict(torch.load(model_path, map_location=DEVICE))


import os
import torch
import numpy as np
import matplotlib.pyplot as plt
from torchvision import transforms
from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay
from pytorch_grad_cam import GradCAM
from pytorch_grad_cam.utils.model_targets import ClassifierOutputTarget
from pytorch_grad_cam.utils.image import show_cam_on_image

# === Setup ===
DEVICE = torch.device("cuda" if torch.cuda.is_available() else "cpu")
OUTPUT_DIR = "/content/drive/MyDrive/Arewa_capstone_project/Summary Charts/vit_villageplant_analysis"
os.makedirs(OUTPUT_DIR, exist_ok=True)

# ===  Confusion Matrix ===
all_preds, all_labels = [], []
model.eval()
with torch.no_grad():
    for imgs, lbls in test_loader:
        imgs, lbls = imgs.to(DEVICE), lbls.to(DEVICE)
        outs = model(imgs)
        _, preds = torch.max(outs, 1)
        all_preds.extend(preds.cpu().numpy())
        all_labels.extend(lbls.cpu().numpy())

cm = confusion_matrix(all_labels, all_preds)
fig_cm, ax = plt.subplots(figsize=(6, 6))
disp = ConfusionMatrixDisplay(confusion_matrix=cm, display_labels=class_names)
disp.plot(ax=ax, cmap="Blues", xticks_rotation=45, colorbar=False)
plt.title("Confusion Matrix - ViT on VillagePlant")
plt.tight_layout()
plt.savefig(os.path.join(OUTPUT_DIR, "confusion_matrix_vit_villageplant.png"))
plt.show()

# ===  Grad-CAM on ViT ===

# Reverse normalization for visualization
inv_normalize = transforms.Normalize(
    mean=[-0.485 / 0.229, -0.456 / 0.224, -0.406 / 0.225],
    std=[1 / 0.229, 1 / 0.224, 1 / 0.225]
)

# --- Define reshape_transform for ViT ---
def vit_reshape_transform(tensor):
    # Remove class token, reshape to (B, H, W, C), then permute to (B, C, H, W)
    # tensor shape: (B, num_patches+1, C)
    tensor = tensor[:, 1:, :]  # remove class token
    h = w = int((tensor.size(1)) ** 0.5)
    return tensor.reshape(tensor.size(0), h, w, tensor.size(2)).permute(0, 3, 1, 2)

# Pick the correct layer (for torchvision.models.vit_b_16, use 'encoder.layers[-1].ln_1')
try:
    target_layers = [model.encoder.layers[-1].ln_1]  # torchvision ViT
except AttributeError:
    target_layers = [model.blocks[-1].norm1]  # HuggingFace ViT

cam = GradCAM(model=model, target_layers=target_layers, reshape_transform=vit_reshape_transform)

# Sample 10 test images
sample_images, sample_labels = next(iter(test_loader))
sample_images, sample_labels = sample_images.to(DEVICE), sample_labels.to(DEVICE)

with torch.no_grad():
    sample_outputs = model(sample_images)
    _, sample_preds = torch.max(sample_outputs, 1)

for i in range(min(10, len(sample_images))):
    input_tensor = sample_images[i].unsqueeze(0)
    rgb_img = inv_normalize(sample_images[i].cpu()).clamp(0, 1)
    rgb_np = rgb_img.permute(1, 2, 0).numpy()

    target = [ClassifierOutputTarget(sample_preds[i].item())]
    grayscale_cam = cam(input_tensor=input_tensor, targets=target)[0]
    cam_image = show_cam_on_image(rgb_np, grayscale_cam, use_rgb=True)

    correct = sample_preds[i] == sample_labels[i]
    color = 'green' if correct else 'red'
    fig, ax = plt.subplots()
    ax.imshow(cam_image)
    ax.set_title(f"True: {class_names[sample_labels[i]]}, Pred: {class_names[sample_preds[i]]}", color=color)
    ax.axis('off')
    plt.savefig(os.path.join(OUTPUT_DIR, f"gradcam_{i}_{class_names[sample_preds[i]]}.png"))
    plt.show()

print(f"✅ Confusion matrix and Grad-CAM plots saved to: {OUTPUT_DIR}")

Reusing cached mean and std

✅ Confusion matrix and Grad-CAM plots saved to: /content/drive/MyDrive/Arewa_capstone_project/Summary Charts/vit_villageplant_analysis

!pip install gradio

Collecting gradio
  Downloading gradio-5.32.0-py3-none-any.whl.metadata (16 kB)
Collecting aiofiles<25.0,>=22.0 (from gradio)
  Downloading aiofiles-24.1.0-py3-none-any.whl.metadata (10 kB)
Requirement already satisfied: anyio<5.0,>=3.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (4.9.0)
Collecting fastapi<1.0,>=0.115.2 (from gradio)
  Downloading fastapi-0.115.12-py3-none-any.whl.metadata (27 kB)
Collecting ffmpy (from gradio)
  Downloading ffmpy-0.6.0-py3-none-any.whl.metadata (2.9 kB)
Collecting gradio-client==1.10.2 (from gradio)
  Downloading gradio_client-1.10.2-py3-none-any.whl.metadata (7.1 kB)
Collecting groovy~=0.1 (from gradio)
  Downloading groovy-0.1.2-py3-none-any.whl.metadata (6.1 kB)
Requirement already satisfied: httpx>=0.24.1 in /usr/local/lib/python3.11/dist-packages (from gradio) (0.28.1)
Requirement already satisfied: huggingface-hub>=0.28.1 in /usr/local/lib/python3.11/dist-packages (from gradio) (0.31.4)
Requirement already satisfied: jinja2<4.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (3.1.6)
Requirement already satisfied: markupsafe<4.0,>=2.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (3.0.2)
Requirement already satisfied: numpy<3.0,>=1.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (2.0.2)
Requirement already satisfied: orjson~=3.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (3.10.18)
Requirement already satisfied: packaging in /usr/local/lib/python3.11/dist-packages (from gradio) (24.2)
Requirement already satisfied: pandas<3.0,>=1.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (2.2.2)
Requirement already satisfied: pillow<12.0,>=8.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (11.2.1)
Requirement already satisfied: pydantic<2.12,>=2.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (2.11.4)
Collecting pydub (from gradio)
  Downloading pydub-0.25.1-py2.py3-none-any.whl.metadata (1.4 kB)
Collecting python-multipart>=0.0.18 (from gradio)
  Downloading python_multipart-0.0.20-py3-none-any.whl.metadata (1.8 kB)
Requirement already satisfied: pyyaml<7.0,>=5.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (6.0.2)
Collecting ruff>=0.9.3 (from gradio)
  Downloading ruff-0.11.12-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl.metadata (25 kB)
Collecting safehttpx<0.2.0,>=0.1.6 (from gradio)
  Downloading safehttpx-0.1.6-py3-none-any.whl.metadata (4.2 kB)
Collecting semantic-version~=2.0 (from gradio)
  Downloading semantic_version-2.10.0-py2.py3-none-any.whl.metadata (9.7 kB)
Collecting starlette<1.0,>=0.40.0 (from gradio)
  Downloading starlette-0.47.0-py3-none-any.whl.metadata (6.2 kB)
Collecting tomlkit<0.14.0,>=0.12.0 (from gradio)
  Downloading tomlkit-0.13.2-py3-none-any.whl.metadata (2.7 kB)
Requirement already satisfied: typer<1.0,>=0.12 in /usr/local/lib/python3.11/dist-packages (from gradio) (0.15.3)
Requirement already satisfied: typing-extensions~=4.0 in /usr/local/lib/python3.11/dist-packages (from gradio) (4.13.2)
Collecting uvicorn>=0.14.0 (from gradio)
  Downloading uvicorn-0.34.3-py3-none-any.whl.metadata (6.5 kB)
Requirement already satisfied: fsspec in /usr/local/lib/python3.11/dist-packages (from gradio-client==1.10.2->gradio) (2025.3.2)
Requirement already satisfied: websockets<16.0,>=10.0 in /usr/local/lib/python3.11/dist-packages (from gradio-client==1.10.2->gradio) (15.0.1)
Requirement already satisfied: idna>=2.8 in /usr/local/lib/python3.11/dist-packages (from anyio<5.0,>=3.0->gradio) (3.10)
Requirement already satisfied: sniffio>=1.1 in /usr/local/lib/python3.11/dist-packages (from anyio<5.0,>=3.0->gradio) (1.3.1)
Collecting starlette<1.0,>=0.40.0 (from gradio)
  Downloading starlette-0.46.2-py3-none-any.whl.metadata (6.2 kB)
Requirement already satisfied: certifi in /usr/local/lib/python3.11/dist-packages (from httpx>=0.24.1->gradio) (2025.4.26)
Requirement already satisfied: httpcore==1.* in /usr/local/lib/python3.11/dist-packages (from httpx>=0.24.1->gradio) (1.0.9)
Requirement already satisfied: h11>=0.16 in /usr/local/lib/python3.11/dist-packages (from httpcore==1.*->httpx>=0.24.1->gradio) (0.16.0)
Requirement already satisfied: filelock in /usr/local/lib/python3.11/dist-packages (from huggingface-hub>=0.28.1->gradio) (3.18.0)
Requirement already satisfied: requests in /usr/local/lib/python3.11/dist-packages (from huggingface-hub>=0.28.1->gradio) (2.32.3)
Requirement already satisfied: tqdm>=4.42.1 in /usr/local/lib/python3.11/dist-packages (from huggingface-hub>=0.28.1->gradio) (4.67.1)
Requirement already satisfied: python-dateutil>=2.8.2 in /usr/local/lib/python3.11/dist-packages (from pandas<3.0,>=1.0->gradio) (2.9.0.post0)
Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.11/dist-packages (from pandas<3.0,>=1.0->gradio) (2025.2)
Requirement already satisfied: tzdata>=2022.7 in /usr/local/lib/python3.11/dist-packages (from pandas<3.0,>=1.0->gradio) (2025.2)
Requirement already satisfied: annotated-types>=0.6.0 in /usr/local/lib/python3.11/dist-packages (from pydantic<2.12,>=2.0->gradio) (0.7.0)
Requirement already satisfied: pydantic-core==2.33.2 in /usr/local/lib/python3.11/dist-packages (from pydantic<2.12,>=2.0->gradio) (2.33.2)
Requirement already satisfied: typing-inspection>=0.4.0 in /usr/local/lib/python3.11/dist-packages (from pydantic<2.12,>=2.0->gradio) (0.4.1)
Requirement already satisfied: click>=8.0.0 in /usr/local/lib/python3.11/dist-packages (from typer<1.0,>=0.12->gradio) (8.2.1)
Requirement already satisfied: shellingham>=1.3.0 in /usr/local/lib/python3.11/dist-packages (from typer<1.0,>=0.12->gradio) (1.5.4)
Requirement already satisfied: rich>=10.11.0 in /usr/local/lib/python3.11/dist-packages (from typer<1.0,>=0.12->gradio) (13.9.4)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.11/dist-packages (from python-dateutil>=2.8.2->pandas<3.0,>=1.0->gradio) (1.17.0)
Requirement already satisfied: markdown-it-py>=2.2.0 in /usr/local/lib/python3.11/dist-packages (from rich>=10.11.0->typer<1.0,>=0.12->gradio) (3.0.0)
Requirement already satisfied: pygments<3.0.0,>=2.13.0 in /usr/local/lib/python3.11/dist-packages (from rich>=10.11.0->typer<1.0,>=0.12->gradio) (2.19.1)
Requirement already satisfied: charset-normalizer<4,>=2 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface-hub>=0.28.1->gradio) (3.4.2)
Requirement already satisfied: urllib3<3,>=1.21.1 in /usr/local/lib/python3.11/dist-packages (from requests->huggingface-hub>=0.28.1->gradio) (2.4.0)
Requirement already satisfied: mdurl~=0.1 in /usr/local/lib/python3.11/dist-packages (from markdown-it-py>=2.2.0->rich>=10.11.0->typer<1.0,>=0.12->gradio) (0.1.2)
Downloading gradio-5.32.0-py3-none-any.whl (54.2 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 54.2/54.2 MB 22.5 MB/s eta 0:00:00
Downloading gradio_client-1.10.2-py3-none-any.whl (323 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 323.3/323.3 kB 28.6 MB/s eta 0:00:00
Downloading aiofiles-24.1.0-py3-none-any.whl (15 kB)
Downloading fastapi-0.115.12-py3-none-any.whl (95 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 95.2/95.2 kB 9.5 MB/s eta 0:00:00
Downloading groovy-0.1.2-py3-none-any.whl (14 kB)
Downloading python_multipart-0.0.20-py3-none-any.whl (24 kB)
Downloading ruff-0.11.12-py3-none-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (11.5 MB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 11.5/11.5 MB 126.9 MB/s eta 0:00:00
Downloading safehttpx-0.1.6-py3-none-any.whl (8.7 kB)
Downloading semantic_version-2.10.0-py2.py3-none-any.whl (15 kB)
Downloading starlette-0.46.2-py3-none-any.whl (72 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 72.0/72.0 kB 7.3 MB/s eta 0:00:00
Downloading tomlkit-0.13.2-py3-none-any.whl (37 kB)
Downloading uvicorn-0.34.3-py3-none-any.whl (62 kB)
   ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ 62.4/62.4 kB 5.7 MB/s eta 0:00:00
Downloading ffmpy-0.6.0-py3-none-any.whl (5.5 kB)
Downloading pydub-0.25.1-py2.py3-none-any.whl (32 kB)
Installing collected packages: pydub, uvicorn, tomlkit, semantic-version, ruff, python-multipart, groovy, ffmpy, aiofiles, starlette, safehttpx, gradio-client, fastapi, gradio
Successfully installed aiofiles-24.1.0 fastapi-0.115.12 ffmpy-0.6.0 gradio-5.32.0 gradio-client-1.10.2 groovy-0.1.2 pydub-0.25.1 python-multipart-0.0.20 ruff-0.11.12 safehttpx-0.1.6 semantic-version-2.10.0 starlette-0.46.2 tomlkit-0.13.2 uvicorn-0.34.3

!python /content/drive/MyDrive/Arewa_capstone_project/App.py

epoch	▁▁▂▂▂▂▃▃▃▄▄▄▅▅▅▅▆▆▆▇▇▇▇██
final_accuracy	▁
final_auc	▁
final_f1	▁
train_acc	▁▄▆▇▇████████████████████
train_loss	█▇▅▄▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
val_acc	▁▅▅█▃▆▅▅▆▆▆▆▆▆▆▅▅▅▅▅▅▅▅▅▅
val_loss	▁▁▁▁▃▂▂▄▄▄▄▅▅▆▇▇█████████

epoch	25
final_accuracy	0.6
final_auc	0.64835
final_f1	0.59596
train_acc	100
train_loss	0.00044
val_acc	60
val_loss	1.42388

epoch	▁▁▂▂▂▂▃▃▃▄▄▄▅▅▅▅▆▆▆▇▇▇▇██
final_accuracy	▁
final_auc	▁
final_f1	▁
train_acc	▁▃▇▇▇████████████████████
train_loss	█▆▅▄▃▂▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
val_acc	▁▄▅▄▅▅██▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇
val_loss	█▆▅▄▃▃▂▁▂▂▂▂▂▂▃▄▄▄▄▃▃▃▃▃▃

epoch	25
final_accuracy	0.91667
final_auc	0.9375
final_f1	0.91608
train_acc	100
train_loss	0.00143
val_acc	91.66667
val_loss	0.32212

epoch	▁▁▂▂▂▂▃▃▃▄▄▄▅▅▅▅▆▆▆▇▇▇▇██
final_accuracy	▁
final_auc	▁
final_f1	▁
train_acc	▁▃▆▇▇████████████████████
train_loss	█▇▆▅▄▃▂▂▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁▁
val_acc	▅▅█▆▆▆▆█▅▃▁▁▁▁▁▃▃▅▅▅▃▃▃▃▃
val_loss	▂▂▁▁▂▂▃▂▄▅▅▆▆▆▆▇▇▇▇▇█████

epoch	25
final_accuracy	0.55
final_auc	0.65934
final_f1	0.53964
train_acc	100
train_loss	0.00077
val_acc	55
val_loss	1.18503

epoch	25
final_accuracy	0.8125
final_auc	0.89286
final_f1	0.81232
train_acc	100
train_loss	0.00052
val_acc	81.25
val_loss	0.6047

epoch	25
final_accuracy	0.83333
final_auc	0.92857
final_f1	0.83333
train_acc	100
train_loss	0.11762
val_acc	83.33333
val_loss	0.3592

epoch	25
final_accuracy	0.75
final_auc	0.79167
final_f1	0.74425
train_acc	100
train_loss	0.09577
val_acc	75
val_loss	0.54562

epoch	25
final_accuracy	0.84375
final_auc	0.94286
final_f1	0.84236
train_acc	99.60938
train_loss	0.07335
val_acc	84.375
val_loss	0.29716

epoch	25
final_accuracy	0.65
final_auc	0.81
final_f1	0.62667
train_acc	100
train_loss	0.00653
val_acc	65
val_loss	0.82813

epoch	25
final_accuracy	0.79688
final_auc	0.85533
final_f1	0.79278
train_acc	99.60938
train_loss	0.04394
val_acc	79.6875
val_loss	0.54712

	Run	Model	Dataset	Accuracy	F1 Score	ROC AUC
0	vit_b_16_Dikumari	vit_b_16	Dikumari	0.600000	0.595960	0.648352
1	vit_b_16_Kasaisa	vit_b_16	Kasaisa	0.916667	0.916084	0.937500
2	vit_b_16_Kukareta	vit_b_16	Kukareta	0.550000	0.539642	0.659341
3	vit_b_16_Combined	vit_b_16	Combined	0.812500	0.812317	0.892857
4	resnet18_Dikumari	resnet18	Dikumari	0.700000	0.696970	0.875000
5	resnet18_Kasaisa	resnet18	Kasaisa	0.833333	0.833333	0.928571
6	resnet18_Kukareta	resnet18	Kukareta	0.750000	0.744246	0.791667
7	resnet18_Combined	resnet18	Combined	0.843750	0.842365	0.942857
8	mobilenet_v3_Dikumari	mobilenet_v3	Dikumari	0.600000	0.583333	0.747475
9	mobilenet_v3_Kasaisa	mobilenet_v3	Kasaisa	0.791667	0.788360	0.859375
10	mobilenet_v3_Kukareta	mobilenet_v3	Kukareta	0.650000	0.626667	0.810000
11	mobilenet_v3_Combined	mobilenet_v3	Combined	0.843750	0.841270	0.899316
12	efficientnet_b0_Dikumari	efficientnet_b0	Dikumari	0.700000	0.670330	0.780220
13	efficientnet_b0_Kasaisa	efficientnet_b0	Kasaisa	0.833333	0.828571	0.953125
14	efficientnet_b0_Kukareta	efficientnet_b0	Kukareta	0.800000	0.761905	0.821429
15	efficientnet_b0_Combined	efficientnet_b0	Combined	0.796875	0.792777	0.855327

	Model	Test Dataset	Accuracy	F1 Score	ROC AUC
0	resnet18	Dikumari	0.514851	0.434852	0.469412
1	resnet18	Kasaisa	0.483607	0.480781	0.503022
2	resnet18	Kukareta	0.582524	0.576780	0.552830
3	resnet18	Combined	0.493750	0.491346	0.502937
4	mobilenet_v3	Dikumari	0.514851	0.511788	0.503137
5	mobilenet_v3	Kasaisa	0.434426	0.427931	0.465110
6	mobilenet_v3	Kukareta	0.533981	0.531818	0.483396
7	mobilenet_v3	Combined	0.521875	0.516086	0.537085
8	efficientnet_b0	Dikumari	0.465347	0.456340	0.432941
9	efficientnet_b0	Kasaisa	0.500000	0.478084	0.588462
10	efficientnet_b0	Kukareta	0.427184	0.427184	0.446415
11	efficientnet_b0	Combined	0.515625	0.501963	0.497024

epoch	▁▂▃▅▆▇█
train_acc	▁▅▆▇███
train_loss	█▆▅▄▂▁▁
val_acc	█▃▁▁▃▁▃
val_loss	▂▁▄▃▄▆█

epoch	▁▂▃▃▄▅▆▆▇█
train_acc	▁▅▇███████
train_loss	█▅▃▂▁▁▁▁▁▁
val_acc	▁█████████
val_loss	█▄▃▁▁▂▄▄▄▄

epoch	▁▂▃▃▄▅▆▆▇█
train_acc	▁▂▅▅▆▇████
train_loss	█▇▆▅▄▃▂▂▁▁
val_acc	▁▃▁█▆▁▆███
val_loss	█▇▆▄▄▆▃▁▁▁

epoch	7
train_acc	98.75
train_loss	0.05548
val_acc	50
val_loss	0.79786

epoch	10
train_acc	98.96907
train_loss	0.00141
val_acc	91.66667
val_loss	0.26997

epoch	10
train_acc	97.56098
train_loss	0.01479
val_acc	100
val_loss	0.07724

epoch	10
train_acc	93.75
train_loss	0.32119
val_acc	60
val_loss	0.68103

epoch	10
train_acc	95.87629
train_loss	0.14986
val_acc	91.66667
val_loss	0.31264

epoch	10
train_acc	95.12195
train_loss	0.2243
val_acc	80
val_loss	0.42321

epoch	10
train_acc	93.35938
train_loss	0.22022
val_acc	81.25
val_loss	0.33969

epoch	▁▂▄▅▇█
train_acc	▁▅▇▇██
train_loss	█▅▃▂▁▁
val_acc	██▁███
val_loss	▁▁▂▄▆█

epoch	6
train_acc	92.68293
train_loss	0.14248
val_acc	50
val_loss	0.81976

epoch	▁▂▃▄▅▅▆▇█
train_acc	▁▅▆▇▇▇███
train_loss	█▅▄▂▂▂▂▁▁
val_acc	▅▅▁█▅▁▁▁▅
val_loss	█▄▅▁▂▄▄▆▆

ArewaDS Deep Learning with PyTorch Cohort 2.0¶

Group 8 ArewaDS Capstone Project¶

First Training Hyper- parameters¶

Models Training¶

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Training Summary¶

Models testing¶

Test results summary and plot¶

Data Regularizaton¶

Hypertuned Model Training¶

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Run history:

Run summary:

Training Summary¶

Models Testing¶

Testing¶

Visual plots of Test losses and Accuracy¶

How Our Best Model (Vit_b_16) Perform on the Combined Tomato Leaf Disease Dataset¶

Visualizing Model Performance and Interpretability: Confusion Matrix and Grad-CAM¶

Model Deployment¶

How Our Best Model `(Vit_b_16)` Perform on the Combined Tomato Leaf Disease Dataset¶

epoch	9
train_acc	99.21875
train_loss	0.0844
val_acc	75
val_loss	0.54159

epoch	10
train_acc	93.90244
train_loss	0.14943
val_acc	90
val_loss	0.41392

epoch	10
train_acc	94.14062
train_loss	0.16704
val_acc	78.125
val_loss	0.37952

	Model	Dataset	Best Val Accuracy (%)
0	vit_b_16	Dikumari	72.727273
1	vit_b_16	Kasaisa	92.307692
2	vit_b_16	Kukareta	81.818182
3	vit_b_16	VillagePlant	68.750000
4	resnet18	Dikumari	72.727273
5	resnet18	Kasaisa	69.230769
6	resnet18	Kukareta	90.909091
7	resnet18	VillagePlant	87.500000
8	mobilenet_v3	Dikumari	54.545455
9	mobilenet_v3	Kasaisa	92.307692
10	mobilenet_v3	Kukareta	72.727273
11	mobilenet_v3	VillagePlant	71.875000
12	efficientnet_b0	Dikumari	81.818182
13	efficientnet_b0	Kasaisa	92.307692
14	efficientnet_b0	Kukareta	90.909091
15	efficientnet_b0	VillagePlant	81.250000